Integrated Model-Centric Decision Support System for Process Industries

Abstract

To bring the advances in modeling, simulation and optimization environments (MSOEs), open-software architectures, and information technology closer to process industries, novel mechanisms and advanced software tools must be devised to simplify the definition of complex model-based problems. Synergistic interactions between complementary model-based software tools must be refined to unlock the potential of model-centric technologies in industries. This dissertation presents the conceptual definition of a single and consistent framework for integrated process decision support (IMCPSS) to facilitate the realistic formulation of related model-based engineering problems. Through the integration of data management, simulation, parameter estimation, data reconciliation, and optimization methods, this framework seeks to extend the viability of model-centric technologies within the industrial workplace. The main contribution is the conceptual definition and implementation of mechanisms to ease the formulation of large-scale data-driven/model-based problems: data model definitions (DMDs), problem formulation objects (PFOs) and process data objects (PDOs). These mechanisms allow the definition of problems in terms of physical variables; to embed plant data seamlessly into model-based problems; and to permit data transfer, re-usability, and synergy among different activities. A second contribution is the design and implementation of the problem definition environment (PDE). The PDE is a robust object-oriented software component that coordinates the problem formulation and the interaction between activities by means of a user-friendly interface. The PDE administers information contained in DMD and coordinates the creation of PFOs and PIFs. Last, this dissertation contributes a systematic integration of data pre-processing and conditioning techniques and MSOEs. The proposed process data management system (pDMS) implements such methodologies. All required manipulations are supervised by the PDE, which represents an important advantage when dealing with high volumes of data. The IMCPSS responds to the need for software tools centered in process engineers for which the complexity of using current modeling environments is a barrier for broader application of model-based activities. Consequently, the IMCPSS represents a valuable tool for process industries, as the facilitation of problem formulation is translated into incorporation of plant data in less error-prone manner, maximization of time dedicated to the analysis of processes, and exploitation of synergy among activities based on process models.